Australian businesses are buying AI hand over fist while sitting on broken data and untrained teams. Filip Ivankovic argues that AI does not fix a business that does not know its own numbers. It just helps it make bad decisions faster and with more confidence.
Nine in ten want the outcome. One in ten can actually do it. That is not a strategy problem. It is a foundations problem.
I have sat in the room on both sides of this. On the agency side, pitching the shiny new layer to a client who could not tell me their own conversion rate. On the client side, watching a leadership team sign off on an AI tool while the customer data underneath it lived in four systems that did not talk to each other. Same problem, two seats. The thing everyone wants to buy is the thing the business is least ready to use.
Here is the thing. AI does not fix a business that does not know its own numbers. It just helps that business make the same bad decisions faster and with more confidence. That is my position and I will hold it. Buying AI before you have clean data is the most expensive way to be wrong at speed.
The tools have never been more available. The foundations and the skill to use them have never been further behind.
We have measured this market, and the plumbing is the problem
Across the Australian businesses we have scored, the weakest dimension by a wide margin is data and tracking. Not brand. Not acquisition. The boring layer underneath everything. The plumbing. The part that decides whether anything you build on top actually works or just looks like it works.
Now look at what the rest of the market is doing. A new report out of Arktic Fox and Six Degrees Executive surveyed more than 100 Australian marketing, digital and ecommerce leaders. 89.1% of retailers said personalisation is strategically important. Only 10% said they have the mature capability to deliver it. So nine in ten want the outcome. One in ten can actually do it. That gap is not a strategy problem. It is a foundations problem.
It gets worse the deeper you go. 73.3% described their data and martech as only partly integrated. More than 80% said their identity resolution and single customer view work is still under development. These are the businesses now spending money on AI-led personalisation. They are buying the engine before they have laid the road.
Sprinkles on a cake nobody baked
This is the bit that gets me. People treat AI like sprinkles. They think, the fundamentals are a bit shaky, but we will put some AI on top and it will all be good. It will not. Everything you put into the cake changes the taste of the cake. If the base is raw, the sprinkles do not save it. They just sit on top of something nobody should be eating.
AI is not a layer you bolt on. It is an ingredient that goes into a business that already runs. Feed it fragmented customer data and broken measurement and it will give you a confident, well-written, beautifully formatted answer that is wrong. Because it sounds so sure of itself, you will act on it faster than you ever would have on your own gut. That is the danger. Not that the AI is dumb. That it makes your guessing feel like knowing.
Share of Australian retailers fully confident their customer and product data foundations are ready for AI-led use cases, even though far more are already buying it. Source: Arktic Fox Inside Digital and eCommerce 2026
The instruments are everywhere. The musicians are not.
There is a second gap and it is just as bad. The skill gap. ADMA surveyed more than a thousand Australian marketers and found 77% use AI at work at least weekly, 52% daily. Three in four leaders say they would not hire someone without AI skills. But only about 3.5% of roughly 13,000 Australian marketing job ads actually ask for the skill. So the want is loud and the training is quiet.
You now have a business handing out a powerful instrument to a team nobody taught to play. Everyone has the violin. Almost nobody can read the music. The businesses I watch fall hardest are the ones who confuse access to the tool with competence at the tool.
This is the same trap I have been banging on about for years. Marketing was never a thinking thing. It was a dreaming thing. Nobody taught marketers to count, so when finance asks what the spend earned, the room goes quiet. AI does not close that gap. It widens it. It lets a team that cannot read a number produce ten times more output, all of it untested, all of it flying blind, all of it delivered with the polish of something that knows what it is doing.
What I would do about it
If you run a business and you are tempted to buy the AI right now, do not. Build the floor first. Here is the order I would work in.
Know your two numbers. Do you know how much money you make? Do you know how much money you spend? We can work back from everything else from there. If you cannot answer those cleanly, no tool on earth fixes it.
Get one clean source of customer truth. Not a buzzword. The plain version. One place where your customer data is correct, current and joined up, so a question about a customer returns the same answer no matter who asks it. Most businesses have the same customer sitting in four systems wearing four different names. Fix that before you ask a machine to personalise anything.
Get your measurement honest. If you turned a channel off and revenue dropped, now you know something. If you turned it off and nothing changed, ask why you were spending there. That logic does not need AI. It needs you to actually capture what is happening, because the digital stuff is different to a physical shopfront. You can eyeball a shopfront or check the cameras. The digital stuff, if you are not capturing it yourself, you lose it. You cannot get it back.
Train the people. Education. The reason the instrument sits idle is nobody was shown how to play it. A team that can read a number and challenge an output is worth more than any subscription. Buy the skill before you buy more software.
Then, and only then, add AI. To a business that can feed it something worth eating. Clean inputs, trained people, honest measurement. Now the machine makes a clever business cleverer instead of making a confused one confused at scale.
AI does not make a guessing business smart. It makes a guessing business fast. Those are very different things.
The window is open, but not for the reason most think
The opportunity here is real and most people are reading it wrong. The advantage over the next two years will not go to whoever buys the most AI. It will go to whoever has the cleanest data and the people who can actually use it. The tools are commodities now. Everyone has the same access. The road, the plumbing and the skill to read your own numbers, that is the part almost nobody is building, which is exactly why it is the part worth building.
Most of the market is about to spend a fortune teaching a machine to make their existing mess look confident. You do not have to be in that group. Bake the cake first. Then add whatever you want on top. It will actually taste like something.
Numbers on the board. That is the lifeblood. Everything else, AI included, is noise until the foundation holds.